Across 64 English authorities and six 2026 scenarios, even the strongest scenario shock was only 13% of the median uncertainty band. In plain English: the model’s assumptions moved the result less than historical forecast error did. The most aggressive challenger surge I could parameterise sits inside the noise the model has produced in past elections. …
# Introduction Every organization loves to call itself “data-driven.” It’s become the gold standard of credibility, the thing you say to shut down dissent in a meeting. But here’s something worth sitting with for a second: the phrase “according to data analytics” can come from two very different places. One is genuine curiosity. The other …
Baptists and Bootleggers: The Hidden Coalition Behind ‘Data-Driven’ Decisions Read More »
AI Agents When building an AI agent, the design choice matters. A single agent may be enough for straightforward tasks, while more complex workflows may need multiple specialised agents working together, with each one responsible for a specific part of the process, such as retrieval, writing, verification, coding, testing or review. This post explains the …
Single Agent vs Multi-Agent: When to Build a Multi-Agent System Read More »
Image by Editor # Introduction Voice-enabled applications are everywhere, from virtual assistants to customer service chatbots. But for developers, building natural-sounding speech into apps has often meant relying on expensive cloud APIs or dealing with robotic, unnatural voices. Mistral AI aims to change that with Voxtral TTS. It is a powerful, open-weight text-to-speech (TTS) model …
Authors: Ahsaas Bajaj and Benjamin S Knight ? We ran 134,400 simulations grounded in real production ML models to find out. The answer depends on what you’re optimizing for, and on a single diagnostic you can compute before fitting a model. If you’ve ever trained a linear model in scikit-learn, you’ve faced this question: RidgeCV, …
Which Regularizer Should You Actually Use? Lessons from 134,400 Simulations Read More »
Image by Editor # Introduction A harsh truth to begin with: textbook data science usually becomes a lie in the real world. Concepts and techniques are taught on finely curated, beautifully bell-curved data variables, but as soon as we venture into the wild of real projects, we are hit with lots of outliers, unduly skewed …
The “Robust” Data Scientist: Winning with Messy Data and Pingouin Read More »
In my first TDS post, I wrote about how to translate a real-world problem into an integer linear program. In my second, I wrote about how to make that program robust against uncertainty. Both were variations on the same idea: take a fuzzy real-world question, squeeze it into an LP, and let a solver do the rest. …
Image by Editor # The Self-Hosted LLM Problem(s) “Run your own large language model (LLM)” is the “just start your own business” of 2026. Sounds like a dream: no API costs, no data leaving your servers, full control over the model. Then you actually do it, and reality starts showing up uninvited. The GPU runs …
Self-Hosted LLMs in the Real World: Limits, Workarounds, and Hard Lessons Read More »
in a situation where you have plenty of ideas on how to improve your product, but no time to test them all? I bet you have. What if I told you that you no longer have to do it all on your own, you can delegate it to AI. It can run dozens (or even …
Image by Author # Introduction Building large language model (LLM) applications is very different from using consumer-facing tools like Claude Code, ChatGPT, or Codex. Those products are great for end users, but when you want to build your own LLM system, you need a lot more control over how everything works behind the scenes. That …
10 Python Libraries for Building LLM Applications Read More »
